[ale] semi [OT] running a robot with linux

[John]

Space Shuttle GPCs?  What would you like to know? I wouldn't call them 
"tiny" since they were modified IBM 360 mainframes, there are 5 on-board 
with at least 3 running and checking each other in a redundant set.  I 
wrote GN&C code for AP101/S computers for 5 years in HAL/S + about 10 
other languages that most people have never heard about.

Those systems each had 108KB of RAM when I started there, but that was 
upgraded to 256KB.  It wasn't about CPU as much as IO channels - except 
during guidance calculations. There were thousands of redundant channels 
for input/output.  I only know about 1 software error that impacted a 
shuttle mission ... and it was during satellite catch, release, catch, 
release, catch, release .... tests.  A few months later, a buddy changed 
that error accumulator from single precision into double precision and 
forced it to be reset to zero at the beginning of every new guidance 
calculation. Fairly simple fix. That code hadn't been touched in 15+ 
years previously.

Lots of stories from my time at JSC working on shuttle, space station 
and in the FCRs.

On 01/24/2011 09:51 PM, Ron Frazier wrote:
> Hopefully, those computers are powerful ENOUGH.  I think the space
> shuttle's original computers were similarly tiny.  Sometimes, more
> complexity breeds more problems.
>
> Ron